Native versions for AllocateFileRange() #2229
Conversation
|
Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/f92555255c2d928ce618d383aa5da46bd195ef43 for binaries and test log. |
|
Test plan: either (1) start syncing from scratch or (2) delete the blocks/rev* files and start a -reindex. In both cases, at any point in time, should the filesize of the blk* files be a multiple of 16 MiB and the rev* files a multiple of 1 MiB. |
|
Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/2d5e0349e19eec9b4c322cdeab37cb0019551fb6 for binaries and test log. |
|
The test succeeded under OS X Version 10.8.2 |
|
Reindex test passes on Fedora 16/17. Needs windows testing. |
|
This is just an optimization, yes? I think it can wait for 0.8.1. |
|
Indeed an optimization, but one that is easy to test. Ok for 0.8.1. |
|
Nit: use #elif ACK |
|
Added a commit to drop the pre-allocated padding when finalizing a block file (=moving to a next one). |
|
ah, good to not waste 30MiB... |
|
@gmaxwell Largest advantage is the ability to cat the files together to create a bootstrap.dat :) ... though I'm sure people would otherwise complain about wasted disk space too :p |
|
Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/1eb578796c98f457546059b66eb98fd0efc8e690 for binaries and test log. |
|
ACK |
Native versions for AllocateFileRange()
Native versions for AllocateFileRange()
7 participants
Implement native versions of AllocateFileRange() for:
This should prevent the double-writing fallback that is used currently.